Comprehensive service system and method for intelligent safety protection

Abstract

The invention discloses a comprehensive service system and method for intelligent safety protection, and relates to the technical field of safety service systems.After state data of a park are intelligently analyzed, potential risk event types are identified, the risk propagation trend is judged through a space-time association algorithm, the risk level of each area in the park is evaluated, and the risk level of each area in the park is evaluated. And generating a risk level map, performing path planning and scheduling optimization on the inspection equipment according to the current available inspection resources and the risk level map, and cooperatively controlling all the inspection equipment to operate based on path planning and scheduling optimization results. The service system generates a high-credibility regional risk level map, takes a current available resource state and the risk level map as input conditions, dynamically generates a routing inspection path and task allocation, effectively improves reasonable utilization and allocation of routing inspection resources, automatically adjusts a scheduling task under the condition that individual routing inspection equipment fails, has sudden obstacles or is limited in path, and improves the routing inspection efficiency. And stable proceeding of the inspection task is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of security service systems, and particularly to an integrated service system and method for intelligent security protection. Background Art

[0002] With the acceleration of the urbanization process, the increase in population density, and the rise in the complexity of critical infrastructure, various security incidents have shown the characteristics of suddenness, diversity, and high risk. Especially in important areas such as industrial parks, transportation hubs, and energy bases, traditional security protection means have been difficult to meet the actual needs, and intelligent upgrading is urgently needed. A service system refers to using advanced information technology, artificial intelligence, big data analysis, Internet of Things, and edge computing and other means to build an intelligent and integrated security protection system covering the entire process of "early warning, prevention and control, response, disposal, and evaluation".

[0003] The existing technologies have the following deficiencies: 1. In the process of traditional regional security protection, the inspection path planning often adopts a fixed route or a simple grid shortest path algorithm, which cannot be dynamically adjusted according to the real-time risk situation of the park. The scheduling of inspection equipment lacks the evaluation of resource priorities. For example, equipment with insufficient power is dispatched to high-intensity areas, or multiple devices are concentrated in low-risk areas, and the resource allocation is seriously unreasonable; 2. There is a lack of unified scheduling logic and weak coordination mechanism among multiple inspection devices, and problems such as task overlap, path conflict, and idle waiting may occur. When a certain device suddenly fails, the system cannot timely adjust other devices to take over its tasks, affecting the overall efficiency.

[0004] Based on this, the present application proposes an integrated service system and method for intelligent security protection, generates a regional risk level map with high credibility, takes the current available resource status and the risk level map as input conditions, dynamically generates inspection paths and task assignments, effectively improves the rational utilization and allocation of inspection resources, and automatically adjusts the scheduling tasks in the case of individual inspection equipment failure, sudden obstacles, or path restrictions, ensuring the stable progress of inspection tasks. Summary of the Invention

[0005] The purpose of the present invention is to provide an integrated service system and method for intelligent security protection to solve the deficiencies in the background art.

[0006] To achieve the above purpose, the present invention provides the following technical solutions: An integrated service method for intelligent security protection, the service method includes the following steps: The monitoring end constructs a multi-source perception network to obtain the park status data in real time; The central control platform identifies the types of potential risk events, uses the spatio-temporal correlation algorithm to judge the risk propagation trend, generates a risk level map after evaluating the risk levels of each area in the park; Perform path planning and scheduling optimization for inspection equipment based on currently available inspection resources and risk level maps, and coordinately control the operation of all inspection equipment based on the path planning and scheduling optimization results.

[0007] Furthermore, the central control platform identifies the type of potential risk events and uses the spatiotemporal correlation algorithm to determine the risk propagation trend, including the following steps: Use data-driven models to analyze the integrated status data in real time to determine whether there are potential risks of abnormal behavior, equipment failure, or environmental changes in the current park, mark the identified risk types, and form a preliminary list of risk events; Match the identified risk events with the spatial structure of the park, determine the starting location of the risk, and analyze the spatial diffusion path of the risk; Considering the diffusion patterns of similar events in historical data, the risk propagation trend is inferred based on the current environment, and the probability and degree of risk impact on different regions in the short term in the future are output.

[0008] Further, after evaluating the risk level of each area in the park, a risk level map is generated, including the following steps: Collect the types, frequency and severity of risk events that have occurred in each region in history, integrate historical risk information with current risk propagation trends, and use intelligent models to assess the current risk level of each region; The park is divided into multiple spatial grids or functional areas, and the risk level results of each area are mapped to the corresponding area location to generate a visual risk level map.

[0009] Furthermore, the historical risk information is integrated with the current risk propagation trend for analysis, and the intelligent model is used to evaluate the current risk level of each area, including the following steps: Calculate the historical risk assessment index of each region based on historical data, obtain the historical risk assessment value of each region, integrate the historical risk assessment value and the environmental adjustment coefficient, and obtain the current risk propagation trend of region i; The intelligent model combines historical risk assessment and current risk propagation trends to evaluate the risk level of each area. The expression is: , where is the risk level of area i, is the current risk propagation trend of region i, is the historical risk assessment value of region i.

[0010] Furthermore, the risk propagation trend is inferred based on the current environment, and the probability and degree of risk impact on different regions in the short term in the future are output, including the following steps: Map the location coordinates of the risk event to the area identifier in the park map to locate the starting area of the risk. The mapping expression is: , where is the starting area of the risk event, is the coordinate of the risk event in the two-dimensional space; Calculate the transition probability of the risk event history spreading from area i to area j. The expression is: , where is the probability of the risk event history spreading from area i to j, is the number of times the risk event history spreads from i to j, is the total number of risk events in area i, and adjust the diffusion probability according to the current environmental factors. The expression is: , where is the dynamic diffusion probability in the current situation, is the historical diffusion probability, is the current environmental adjustment coefficient; Use the Markov propagation model to calculate the probability of each area being affected at the next moment. The Markov propagation model expression is: , where is the risk probability vector of each area at present, is the adjusted area transition matrix, and the elements of the area transition matrix are , is the risk probability of each area predicted for the next time step; The calculation formula for the risk impact degree of each area is: , where is the risk impact degree of area j, is the affected probability of area j, is the importance assignment of area j.

[0011] Furthermore, mark the identified risk types to form a preliminary risk event list, including the following steps: Extract features from the fused multi-source status data to construct a status representation vector. The expression is: , where represents the status feature vector at the current moment t, represents the i-th feature item, and n is the total number of extracted feature dimensions; Calculate the deviation degree between the current status vector and the historical normal status. The expression is: , where is the status anomaly score at the current moment t, represents the status feature vector at the current moment t, is the mean feature vector of the historical normal status, represents the Euclidean distance; Using the trained decision tree classification model, identify the risk event type for the current state. The identification algorithm expression is: , where is the identified risk type label, is the trained decision tree classification model, represents the state feature vector at the current moment t; Organize the detected event type, occurrence time, spatial location, and model confidence to form a structured risk event list.

[0012] Furthermore, fuse the historical risk assessment value and the environmental adjustment coefficient to obtain the current risk propagation trend of region i. The expression is: , where is the current risk propagation trend of region i, is the dynamic diffusion probability in the current situation, is the historical risk assessment value of region i, is the current environmental adjustment coefficient.

[0013] Furthermore, the monitoring end constructs a multi-source perception network to obtain the park status data in real time, including the following steps: The monitoring end obtains the information data of all perception devices and inspection devices deployed in the current park protection area through the API interface provided by the protection area management platform, including device type, geographical location, operating status, device number, and communication protocol; Based on the obtained perception device information, organize and manage it according to functional categories and spatial locations, and construct a multi-source perception network composed of various types of sensors such as camera monitoring, infrared detection, temperature and humidity sensing, and access control systems; Real-time collect various types of status data in the park through various types of sensor devices in the perception network. The status data includes environmental data, personnel data, and device status data.

[0014] Furthermore, constructing a multi-source perception network composed of various types of sensors such as camera monitoring, infrared detection, temperature and humidity sensing, and access control systems includes the following steps: Classify and manage the standardized device information according to functional types to form multiple types of perception sources. According to the device geographical location information, perform spatial grid division to form the distribution of perception devices within each grid. The expression is: , where represents the spatial coordinates of the device, represents the grid number of the i-th row and j-th column in the two-dimensional space division area; Combine the device functional type and spatial location to form a joint mapping relationship of perception devices under multiple types and multiple regions, and construct a multi-source perception network. The function expression is: , where represents a set of devices of type ; represents the grid number of the i-th row and j-th column of the area after two-dimensional space division, represents under the type and the grid conditions; Output the hierarchical structure as a multi-source perception network topology for subsequent state collection and collaborative analysis.

[0015] This application also provides an integrated service system for intelligent security protection, including a data perception module, a risk identification module, and a collaborative control module; Data perception module: Obtain all perception devices and patrol device information in the protection area through the API interface of the protection area management platform, construct a multi-source perception network with the perception devices, and obtain the park status data in real time; Risk identification module: After intelligently analyzing the park status data, identify potential risk event types, use the spatio-temporal correlation algorithm to judge the risk propagation trend, use the Transformer model to evaluate the risk levels of each area in the park, and generate a risk level map; Collaborative control module: Plan the paths and optimize the scheduling of the patrol devices according to the current available patrol resources and the risk level map, and collaboratively control the operation of all patrol devices based on the path planning and scheduling optimization results.

[0016] In the above technical solution, the technical effects and advantages provided by the present invention are: 1. After intelligently analyzing the park status data, the present invention identifies potential risk event types, uses the spatio-temporal correlation algorithm to judge the risk propagation trend, uses the Transformer model to evaluate the risk levels of each area in the park, and generates a risk level map. According to the current available patrol resources and the risk level map, plan the paths and optimize the scheduling of the patrol devices, and collaboratively control the operation of all patrol devices based on the path planning and scheduling optimization results. This service system generates a regional risk level map with high credibility, takes the current available resource status and the risk level map as input conditions, dynamically generates patrol paths and task assignments, effectively improves the rational utilization and allocation of patrol resources, and automatically adjusts the scheduling tasks in the case of individual patrol device failures, sudden obstacles or path restrictions, ensuring the stable progress of patrol tasks.

[0017] 2. The present invention obtains all the sensing devices and inspection device information in the protection area through the API interface of the protection area management platform, constructs a multi-source sensing network for the sensing devices, and obtains the park status data in real time. It obtains all the sensing devices and inspection devices through the interface of the protection area management platform, realizes the integrated management of multi-source heterogeneous devices, uses the edge server for preprocessing, reduces the central load, realizes low-latency data reporting, and multiple devices cooperate to form a seamlessly covered "digital sensing skin", providing a data basis for grasping the park situation. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required to be used in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained according to these drawings.

[0019] Figure 1 It is a mind map of the service method of the present invention.

[0020] Figure 2 It is a system architecture diagram of the service system of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0022] Embodiment 1: Please refer to Figure 1 As shown, the comprehensive service method for intelligent security protection in this embodiment includes the following steps: The monitoring terminal obtains all the sensing devices and inspection device information in the protected area through the API interface of the protected area management platform, constructs a multi-source sensing network with the sensing devices, and obtains the park status data in real time. All the sensing devices and inspection devices perform preliminary preprocessing on the data through the edge server and upload it to the central control platform. After the central control platform intelligently analyzes the park status data, it identifies potential risk event types, uses a spatio-temporal correlation algorithm to judge the risk propagation trend, uses a Transformer model to evaluate the risk levels of each area in the park (evaluated by combining the current risk propagation trend and historical data of the area), and generates a risk level map. According to the current available inspection resources (device location, power, task status) and the risk level map, path planning and scheduling optimization are carried out for the inspection devices, and based on the results of path planning and scheduling optimization, all inspection devices are controlled to run in a coordinated manner. After the inspection is completed, the devices report the process data and detected abnormalities to the central control platform.

[0023] After intelligently analyzing the park status data in this application, potential risk event types are identified, a spatio-temporal correlation algorithm is used to judge the risk propagation trend, a Transformer model is used to evaluate the risk levels of each area in the park, and a risk level map is generated. According to the current available inspection resources and the risk level map, path planning and scheduling optimization are carried out for the inspection devices, and based on the results of path planning and scheduling optimization, all inspection devices are controlled to run in a coordinated manner. This service system generates a regional risk level map with high credibility, takes the current available resource status and the risk level map as input conditions, dynamically generates inspection paths and task assignments, effectively improves the rational utilization and allocation of inspection resources, and automatically adjusts the scheduling tasks in the case of individual inspection device failures, sudden obstacles or path limitations, ensuring the stable progress of inspection tasks.

[0024] This application obtains all the sensing devices and inspection device information in the protected area through the API interface of the protected area management platform, constructs a multi-source sensing network with the sensing devices, obtains the park status data in real time, obtains all the sensing devices and inspection devices through the interface of the protected area management platform, realizes the integrated management of multi-source heterogeneous devices, uses the edge server for preprocessing, reduces the central load, realizes low-latency data reporting, and multiple devices cooperate to form a seamless coverage "digital sensing skin", providing a data basis for grasping the park situation.

[0025] Embodiment 2: Please refer to Figure 2 As shown, the integrated service system for intelligent security protection in this embodiment includes a data sensing module, a risk identification module, and a collaborative control module; Data Sensing Module: Obtain all the sensing devices and inspection device information in the protected area through the API interface of the protected area management platform, construct a multi-source sensing network with the sensing devices, and obtain the park status data in real time. All the sensing devices and inspection devices perform preliminary preprocessing on the data through the edge server, and the park status data is sent to the risk identification module; Risk Identification Module: After intelligently parsing the park status data, identify the types of potential risk events, use the spatio-temporal correlation algorithm to judge the risk propagation trend, use the Transformer model to evaluate the risk levels of each area in the park (evaluated by combining the current risk propagation trend and historical data of the area), and generate a risk level map. The risk level map is sent to the collaborative control module; Collaborative Control Module: Based on the current available inspection resources (device location, power, task status) and the risk level map, perform path planning and scheduling optimization for the inspection devices, and collaboratively control the operation of all inspection devices based on the results of path planning and scheduling optimization.

[0026] Embodiment 3: The monitoring end obtains all the sensing devices and inspection device information in the protected area through the API interface of the protected area management platform, constructs a multi-source sensing network with the sensing devices, obtains the park status data in real time, and all the sensing devices and inspection devices perform preliminary preprocessing on the data through the edge server and upload it to the central control platform, including the following steps: The monitoring end obtains the information data of all the sensing devices and inspection devices deployed in the current park protected area through the standard API interface provided by the protected area management platform, including basic attributes such as device type, geographical location, operating status, device number, communication protocol, etc.

[0027] Based on the obtained sensing device information, organize and manage it according to functional categories and spatial positions, and construct a multi-source sensing network composed of various types of sensors such as camera monitoring, infrared detection, temperature and humidity sensing, access control systems, etc., to achieve full coverage and perception of the park environment status.

[0028] Through the continuous operation of various types of sensor devices in the sensing network, various status data in the park are collected in real time, including environmental data (such as temperature, humidity, smoke concentration), personnel data (such as abnormal intrusion, abnormal trajectory), device status data (such as power, fault), etc.

[0029] All the sensing devices and inspection devices first transmit the real-time collected data to the edge server deployed locally, and the edge computing node performs preliminary preprocessing on the original data, including data denoising, format standardization, abnormal data marking, event compression, etc., to reduce the central computing pressure; the preprocessed data is uniformly uploaded to the park central control platform for in-depth analysis.

[0030] Construct a multi-source perception network composed of various types of sensors such as camera monitoring, infrared detection, temperature and humidity sensing, and access control systems, including the following steps: Standardize the device information of camera monitoring, infrared detection, temperature and humidity sensing, and access control systems; classify and manage the standardized device information according to functional types to form multiple types of perception sources, and divide the space into grids according to the device geographical location information to form the distribution of perception devices in each grid. The expression is: , where represents the spatial coordinates of the device, represents the grid number of the i-th row and j-th column in the area after two-dimensional space division, The function maps the coordinate position to the discrete space grid, which is used to establish the relationship between the devices in the area, combines the device functional type and spatial position, forms the joint mapping relationship of the perception devices under multiple types and multiple regions, and constructs a multi-source perception network. The function expression is: , where represents the device set of type , represents the grid number of the i-th row and j-th column in the area after two-dimensional space division, represents the device set under the conditions of type and grid . This function is used to establish a structured distribution network of perception devices in the "functional type - spatial position" dual dimension, output the hierarchical structure as the topology of the multi-source perception network, and be used for subsequent state acquisition and collaborative analysis. The function expression is: , where represents the device set under the conditions of type and grid , represents the topology structure of the constructed multi-source perception network. This function is used to integrate the device organizational structures under all functional types and spatial regions to form the general map of the multi-source perception network for the service system, traverse and merge all the classified and mapped , and the result can be structured data (such as graph structure or nested dictionary).

[0031] In this application, the perception device information obtained from the API interface is field-normalized to unify the data structure to ensure the consistency of subsequent processing. Example field standardization code (Python): def_standardize_device_data(raw_device_data): return{ "device_id":raw_device_data.get("id"), "device_type": raw_device_data.get("type"), # e.g., camera, infrared "location": raw_device_data.get("coordinates"), # (x,y) "status": raw_device_data.get("status"), "zone_id": raw_device_data.get("zone_id"), "last_updated": raw_device_data.get("timestamp") } Classify the sensing devices by functional type, such as camera monitoring, infrared detection, temperature and humidity sensing, access control system, etc. Classification code (create an index by type): from collections import defaultdict def categorize_by_type(devices): category_map = defaultdict(list) for device in devices: category_map[device["device_type"]].append(device) return category_map According to the device location information, classify the devices into different regional grids or functional partitions. Example code for grid-based regional division: def assign_device_to_grid(device, grid_size = 10): x, y = device["location"] grid_x = int(x / grid_size) grid_y = int(y / grid_size) return f"grid_{grid_x}_{grid_y}" Example of the structure after assignment to the grid: # Device distribution table { "grid_0_1":[device_1,device_5], "grid_1_0":[device_2], "grid_1_1":[device_3,device_4] } Integrate type classification and spatial location mapping to build a structured multi-source perception network. Network structure model (nested dictionary): def_build_multi_source_network(categorized_devices): multi_source_network=defaultdict(lambda:defaultdict(list)) for_device_type,device_listincategorized_devices.items(): for_device_in_device_list: grid_id=assign_device_to_grid(device) multi_source_network[device_type][grid_id].append(device) return_multi_source_network The multi-source perception network data table is shown in Table 1: Multi-source perception network data table Table 1 Based on the acquired perception device information, it is organized and managed according to functional categories and spatial locations, and a multi-source perception network consisting of various types of sensors such as video surveillance, infrared detection, temperature and humidity sensing, and access control systems is constructed.

[0032] After the central control platform intelligently analyzes the park status data, it identifies the type of potential risk events, uses the spatiotemporal correlation algorithm to determine the risk propagation trend, uses the Transformer model to evaluate the risk level of each area in the park (combined with the current risk propagation trend of the area and historical data evaluation), and generates a risk level map, including the following steps: Use data-driven models to analyze the fused status data in real time to determine whether there are potential risks such as abnormal behavior, equipment failure, or environmental changes in the current park, mark the identified risk types, and form a preliminary list of risk events (such as fire, illegal intrusion, equipment overheating, etc.).

[0033] Match the identified risk events with the spatial structure of the park, determine the starting position of the risk, analyze the spatial diffusion path of the risk, such as the spread of a fire to the upwind direction and the spread of abnormal behavior to adjacent areas. Consider the diffusion patterns of similar events in historical data, and infer the risk propagation trend in combination with the current environment (wind speed, temperature, pedestrian flow distribution, etc.), and output the probability and degree of possible risk impact on different regions in the short term in the future.

[0034] Collect the types, frequencies, and severities of risk events that have occurred in each region historically, conduct a fusion analysis of the historical risk information and the current risk propagation trend, and use an intelligent model (such as a deep learning model based on the attention mechanism) to evaluate the risk level of each region currently, which is usually divided into multiple levels such as high risk, medium risk, and low risk.

[0035] Divide the park into multiple spatial grids or functional areas, map the risk level results of each region to the corresponding regional positions, generate a visual risk level map, which can be displayed in the form of a heat map, a two-dimensional plan, or a three-dimensional park model in this application. Synchronize the map to the decision-making terminal and the dispatching system to provide a basis for subsequent inspection path planning and response dispatching.

[0036] Use a data-driven model to analyze the fused status data in real time, judge whether there are potential risks of abnormal behavior, equipment failure, or environmental change in the current park, mark the identified risk types, and form a preliminary list of risk events, including the following steps: Extract key features from the fused multi-source status data to construct a unified status representation vector, and the expression is: , where represents the status feature vector at the current moment t, represents the i-th feature item (such as the pedestrian flow density of the camera, the infrared heat, the access card swiping frequency, the temperature and humidity value, etc.), n is the total number of extracted feature dimensions, and constructing a unified status representation vector is used to uniformly model the status of the park at the current moment, providing an input basis for subsequent risk judgment.

[0037] Calculate the deviation degree between the current status vector and the historical normal status to identify whether the whole is abnormal, and the expression is: , where is the status anomaly score at the current moment t, represents the status feature vector at the current moment t, is the mean feature vector of the historical normal status, represents the Euclidean distance (L2 norm). By measuring the deviation degree between the current status and the "normal status", the larger the value, the more abnormal the status, which is convenient to be used as a preliminary screening basis for anomaly detection.

[0038] Use the trained decision tree classification model to identify the type of risk event for the current state. The recognition algorithm expression is as follows: , where is the recognized risk type label, is the trained decision tree classification model, represents the state feature vector at the current moment t. By inputting the current state vector into the classification model, the corresponding risk event type label is output to achieve the identification of specific risk events.

[0039] Organize the detected event type, occurrence time, spatial location, and model confidence to form a structured risk event list. The data structure is shown as follows: Risk event list = { Time: t, Location: Area A, Type: Fire, Confidence: 0.92 }, { Time: t, Location: Area B, Type: Intrusion, Confidence: 0.88 }, The risk event list is used for subsequent risk propagation analysis, scheduling optimization, and platform visualization display.

[0040] Match the identified risk events with the spatial structure of the park to determine the starting position of the risk, analyze the diffusion path of the risk in space, consider the diffusion patterns of similar events in historical data, and infer the risk propagation trend based on the current environment. Output the probability and degree of risk impact on different regions in the short term in the future, including the following steps: Map the position coordinates of the risk event to the area identifier in the park map to locate the starting area of the risk. The mapping expression is as follows: , where is the starting area of the risk event, is the coordinate of the risk event in the two-dimensional space, is the area mapping function (determine which area a point belongs to according to the park division rules). By clarifying the spatial starting point of the risk event, it is a key prerequisite for subsequent diffusion modeling.

[0041] ​Use the park layout map to construct a connected graph structure between regions, representing the risk reachable paths, which is the basic structure for representing the possibility of risk spreading from one region to another. Calculate the transfer probability of the risk event history spreading from region i to region j. The expression is: , where is the probability that the risk event history spreads from region i to j, is the number of times the risk event history spreads from i to j, is the total number of risk events in region i, and adjust the spread probability according to the current environmental factors. The expression is: , where is the dynamic spread probability in the current situation, is the historical spread probability, is the current environmental adjustment coefficient.

[0042] Use the Markov propagation model to calculate the probability that each region will be affected at the next moment. The Markov propagation model expression is: , where is the current risk probability vector of each region, is the adjusted regional transfer matrix, and the elements of the regional transfer matrix are , is the risk probability of each region predicted for the next time step. By simulating the spread of risk in the park according to a certain rule, the probability values that each region may be affected in the short term are obtained. Then the calculation formula for the risk impact degree of each region is: , where is the risk impact degree of region j, is the probability of region j being affected, is the importance assignment of region j. The importance assignment of the region is obtained by adding the normalized value of the pedestrian flow density in the region and the normalized value of the number of risk events occurring.

[0043] The calculation expression of the current environmental adjustment coefficient is: , where is the wind direction assignment (judge the influence of region i on region j. If it is downwind (i.e., the wind direction blows from region i to region j), it is 1; if it is upwind (i.e., the wind direction blows from region j to region i), it is 0; if it is crosswind (the wind direction forms an angle of 80° to 100° or 260° to 280° with the connection line between regions i and j), it is 0.5)), is the wind speed, is the pedestrian flow density, , are the weight coefficients, and .

[0044] The logic for obtaining the adjusted regional transfer matrix is as follows: By analyzing the historical risk event propagation paths, statistically calculating the probability of spreading from region i to region j, constructing an initial transfer matrix, integrating historical and environmental information, and calculating the adjusted transfer matrix, the expression is: , where is the adjusted regional transfer matrix, is the dynamic diffusion probability in the current situation, and m is the number of regions.

[0045] Collect the types, frequencies, and severities of risk events that have occurred in each region historically, conduct a comprehensive analysis by integrating historical risk information with the current risk propagation trend, and use an intelligent model to evaluate the risk level of each region currently, including the following steps: Calculate the historical risk assessment indicators for each region based on historical data. The historical risk assessment indicators usually consider the frequency and severity of risk events, and a weighted average method can be used to calculate the historical risk assessment value for each region. The expression is: , where is the historical risk assessment value of region i, is the normalized value of the frequency of risk events that occurred in region i during the historical period, is the normalized value of the impact range of historical risk events in region i (i.e., the area of the region affected when the risk event occurred), is the total duration of the historical period, is the maximum value of the impact range, , are the weight coefficients, and .

[0046] The assessment of the current risk propagation trend considers the transfer probability between regions and the propagation dynamics of current risk events. By integrating the historical risk assessment value and the environmental adjustment coefficient, the current risk propagation trend of region i is obtained. The expression is: , where is the current risk propagation trend of region i, is the dynamic diffusion probability in the current situation, is the historical risk assessment value of region i, is the current environmental adjustment coefficient.

[0047] Through an intelligent model (such as Transformer), combining the historical risk assessment and the current risk propagation trend, the risk level of each region is evaluated. Assuming that Transformer is used for spatio-temporal correlation modeling, the risk level of region i is obtained. The expression is: , where is the risk level of region i, An intelligent model based on historical and propagation trends for evaluating the current risk level.

[0048] The Transformer model is a deep learning model commonly used to process time series data. Its core advantage lies in its ability to capture dependencies in long time series through the self-attention mechanism, especially suitable for processing complex spatio-temporal data and tasks with multi-level context relationships. In the task of evaluating the current risk level, the Transformer can dynamically generate risk levels for each region based on historical data and current risk propagation trend information.

[0049] The following is the processing logic for evaluating the current risk level based on the Transformer model: First, historical risk data and the current risk propagation trend need to be used as the input to the Transformer model. The format of the input data usually includes the following parts: historical risk assessment values for each region, which include historical data such as the frequency and severity of risks, and the current risk propagation trend for each region, which combines information such as the risk propagation dynamics of other regions and environmental factors. The historical risk data is divided by time steps to construct time series data. Based on the relative risk propagation relationship between current regions, the risk propagation trend can be input by time step.

[0050] The core of the Transformer model is the self-attention mechanism, which allows the model to consider information from other positions when processing the input at a certain position. When evaluating the current risk level, the model needs to comprehensively evaluate the risks of each region based on historical risks and the current propagation trend. Therefore, it is necessary to learn the relationships between regions and between time steps through the self-attention mechanism.

[0051] To enable the model to focus on the features of different subspaces, the Transformer uses the multi-head self-attention mechanism. The multi-head self-attention maps the input data to multiple subspaces, calculates different attention weights through multiple self-attention heads in parallel, and finally concatenates and linearly transforms these results.

[0052] Each layer of the Transformer also contains a feed-forward neural network for further processing of each input. This network usually contains two linear transformations and an activation function (such as ReLU).

[0053] Since Transformer does not have built-in temporal or spatial information, positional encoding is needed to supplement the positional information of sequential data. Positional encoding enables the model to understand the sequential information in the data, which is particularly important in spatio-temporal sequence modeling. After being processed by multiple layers of self-attention and feed-forward neural networks, the final output of Transformer is the risk score for each region. These scores represent the current risk level of the region.

[0054] Through the above processing steps, the Transformer model combines information such as historical risk data, current propagation trends, and environmental factors to output the current risk level for each region. Through this assessment, the risk situation of each region can be obtained and a risk level map can be generated.

[0055] Based on the current available inspection resources (equipment location, power, task status) and the risk level map, path planning and scheduling optimization are carried out for the inspection equipment. Based on the results of path planning and scheduling optimization, all inspection equipment operations are coordinated and controlled. After the inspection is completed, the equipment reports the process data and detected anomalies to the central control platform, including the following steps: Collect the status information of all inspection equipment, including: Equipment location: The current location of each inspection equipment within the park.

[0056] Power status: The remaining power of each equipment.

[0057] Task status: The current working task status of each equipment (such as idle, in progress, on standby, etc.).

[0058] Input data: Equipment location: , where are the position coordinates of the i-th inspection equipment; Power status: , is the power percentage of the i-th equipment; Task status: , represents the task status of the i-th equipment, and M represents the number of equipment.

[0059] According to the risk level map generated by the central control platform through intelligent analysis, the risk level of each region is evaluated based on historical data and risk propagation trends, and the risk level of each region is output. Risk level map: , where, is the risk level of the i-th region, indicating the severity of the risk in that region.

[0060] Based on the location, power, and task status of the inspection equipment, as well as the risk level map, path planning and scheduling optimization are carried out, mainly including the following contents: Device task allocation: According to the idle state of the device, the level of risk, and the battery status, high-risk areas are preferentially allocated to the inspection devices with sufficient battery power.

[0061] Path planning: Based on the current location of the device and the risk level of the target area, the best inspection path is planned. Path planning needs to consider the device's battery power and task status to ensure that the task is completed before the device runs out of power.

[0062] Scheduling optimization: Under the constraints of the device's location, battery power, and task status, the scheduling order of the devices is optimized to maximize the inspection efficiency of the park.

[0063] For each inspection device, if indicates that the device is in an idle state and the battery is greater than the battery threshold, the device is assigned to an area with a higher risk level (i.e., the high-risk area , where Z2 represents the risk threshold), and tasks are preferentially assigned to the high-risk area.

[0064] Based on the device's battery status, task status, and current workload, scheduling optimization algorithms (such as genetic algorithms, particle swarm algorithms, etc.) are used to optimize the scheduling order of the devices to ensure that tasks can be completed efficiently and the device's battery consumption is minimized. Using scheduling optimization algorithms to optimize the scheduling order of the devices belongs to the prior art and will not be elaborated in this application.

[0065] Classic path planning algorithms (such as A* algorithm, Dijkstra algorithm) are used to calculate the shortest path from the current location of each inspection device to the target area. This step belongs to the prior art and will not be elaborated in this application.

[0066] Based on path planning and scheduling optimization, all inspection devices are controlled in real-time and collaboratively. Ensure that the devices move according to the path planning and complete the tasks within the specified time.

[0067] The status of the inspection devices is monitored in real-time, and the path is adjusted to handle emergencies (such as insufficient battery power, device failure, etc.). Coordinate the actions of different devices to avoid path conflicts and task overlaps. When the inspection device completes the task, the device will upload the data during the inspection process (such as device status, inspection results, detected abnormalities) to the central control platform for data aggregation and exception handling. Process data includes the device's inspection path, execution time, status information, etc. Exception reports include device failures, environmental changes, potential risks, etc. detected during the inspection process.

[0068] The central control platform aggregates the data reported by all the inspected devices, analyzes the inspection results, and confirms the operation status of the devices and the completion status of the tasks. For the abnormal events found (such as equipment failures, risk events, etc.), the central control platform adjusts the inspection task arrangements according to the data feedback and reallocates the inspection device tasks. The subsequent path planning and scheduling optimization strategies are adjusted according to the results of this inspection to ensure the more efficient completion of future inspection tasks.

[0069] It should be understood that the term "and / or" in this document is merely a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. Here, A and B can be singular or plural. In addition, the character " / " in this document generally represents an "or" relationship between the associated objects before and after, but it may also represent an "and / or" relationship, which can be specifically understood by referring to the context before and after.

[0070] It should be understood that in various embodiments of the present application, the magnitudes of the serial numbers of the above processes do not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0071] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed in this document can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0072] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in the present application, and all should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A comprehensive service method for intelligent security protection, characterized by: The service method comprises the following steps: The monitoring end builds a multi-source perception network to obtain park status data in real time; The central control platform identifies the types of potential risk events, uses spatiotemporal correlation algorithms to determine the risk propagation trend, integrates historical risk information with current risk propagation trends, and uses intelligent models to assess the current risk level of each region; Map the risk level results of each area to the corresponding area location to generate a visual risk level map; Perform path planning and scheduling optimization for inspection equipment based on currently available inspection resources and risk level maps, and coordinately control the operation of all inspection equipment based on the path planning and scheduling optimization results.

2. The comprehensive service method for intelligent security protection according to claim 1, characterized in that: The central control platform identifies the types of potential risk events and uses the spatiotemporal correlation algorithm to determine the risk propagation trend, including the following steps: Use data-driven models to analyze the integrated status data in real time to determine whether there are potential risks of abnormal behavior, equipment failure, or environmental changes in the current park, mark the identified risk types, and form a preliminary list of risk events; Match the identified risk events with the spatial structure of the park, determine the starting location of the risk, and analyze the spatial diffusion path of the risk; Considering the diffusion patterns of similar events in historical data, the risk propagation trend is inferred based on the current environment, and the probability and degree of risk impact on different regions in the short term in the future are output.

3. The comprehensive service method for intelligent security protection according to claim 2, characterized in that: After assessing the risk level of each area in the park, a risk level map is generated, including the following steps: Collect the types, frequency and severity of risk events that have occurred in each region in history, integrate historical risk information with current risk propagation trends, and use intelligent models to assess the current risk level of each region; The park is divided into multiple spatial grids or functional areas, and the risk level results of each area are mapped to the corresponding area location to generate a visual risk level map.

4. The comprehensive service method for intelligent security protection according to claim 3 is characterized in that: The historical risk information is integrated with the current risk propagation trend for analysis, and the intelligent model is used to assess the current risk level of each area, including the following steps: Calculate the historical risk assessment index of each region based on historical data, obtain the historical risk assessment value of each region, integrate the historical risk assessment value and the environmental adjustment coefficient, and obtain the current risk propagation trend of region i; The intelligent model combines historical risk assessment and current risk propagation trends to evaluate the risk level of each area. The expression is: , where is the risk level of area i, is the current risk propagation trend of region i, is the historical risk assessment value of region i.

5. The comprehensive service method for intelligent security protection according to claim 3 is characterized in that: Infer the risk propagation trend based on the current environment and output the probability and degree of risk impact on different regions in the short term in the future, including the following steps: Map the location coordinates of the risk event to the area identifier in the park map to locate the risk starting area. The mapping expression is: , where is the starting area of ​​the risk event, is the coordinate of the risk event in two-dimensional space; Calculate the transfer probability of risk event history from region i to region j, the expression is: , where is the probability that the risk event history spreads from region i to region j, is the number of times the risk event history spreads from i to j, is the total number of risk events in region i, and the diffusion probability is adjusted according to the current environmental factors. The expression is: , where is the dynamic diffusion probability under the current situation, is the historical diffusion probability, is the current environment adjustment factor; Use the Markov propagation model to calculate the probability of each area being affected at the next moment. The expression of the Markov propagation model is: , where is the risk probability vector of each region at present, is the adjusted regional transfer matrix, and the regional transfer matrix elements are , To predict the risk probability of each region in the next time step; The formula for calculating the risk impact of each region is: , where is the risk impact degree of region j, is the probability of area j being affected, Assign a value to the importance of region j.

6. The comprehensive service method for intelligent security protection according to claim 3 is characterized in that: Mark the identified risk types and form a preliminary risk event list, including the following steps: Extract features from the fused multi-source state data and construct a state representation vector, which is expressed as: , where represents the state feature vector at the current time t, represents the i-th feature item, and n is the total number of feature dimensions extracted; Calculate the degree of deviation between the current state vector and the historical normal state. The expression is: , where is the abnormal state score at the current time t, represents the state feature vector at the current time t, is the mean eigenvector of the historical normal state, represents the Euclidean distance; Use the trained decision tree classification model to identify the risk event type of the current state. The identification algorithm expression is: , where Label the identified risk type. is the trained decision tree classification model, Represents the state feature vector at the current time t; The detected event types, occurrence times, spatial locations and model confidence levels are collated to form a structured risk event list.

7. The comprehensive service method for intelligent security protection according to claim 6, characterized in that: By integrating the historical risk assessment value and the environmental adjustment coefficient, the current risk propagation trend of region i is obtained, which is expressed as: , where is the current risk propagation trend of region i, is the dynamic diffusion probability under the current situation, is the historical risk assessment value of region i, is the current environment adjustment factor.

8. The comprehensive service method for intelligent security protection according to claim 7, characterized in that: The monitoring end builds a multi-source perception network to obtain park status data in real time, including the following steps: The monitoring end obtains the information data of all sensing devices and inspection devices deployed in the current park protection area through the API interface provided by the protection area management platform, including device type, geographical location, operating status, device number, and communication protocol; Based on the acquired sensing device information, it is organized and managed according to functional categories and spatial locations, and a multi-source sensing network consisting of various types of sensors such as video surveillance, infrared detection, temperature and humidity sensing, and access control systems is constructed; Through various types of sensor devices in the perception network, various status data in the park are collected in real time. The status data includes environmental data, personnel data and equipment status data.

9. The comprehensive service method for intelligent security protection according to claim 8, characterized in that: Building a multi-source perception network consisting of multiple types of sensors such as video surveillance, infrared detection, temperature and humidity sensing, and access control systems includes the following steps: The standardized device information is classified and managed according to the functional type to form multiple types of perception sources. The spatial grid is divided according to the device geographical location information to form the distribution of perception devices in each grid. The expression is: , where Represents the spatial coordinates of the device, Represents the grid number of the i-th row and j-th column of the area after the two-dimensional space is divided; The device function type and spatial location are combined to form a joint mapping relationship between sensing devices in multiple types and multiple regions, and a multi-source sensing network is constructed. The function expression is: , where Indicates the type A collection of devices, It represents the grid number of row i and column j in the area after the two-dimensional space is divided. Indicated in type and Grid Equipment collection under conditions; The hierarchical structure is output as a multi-source perception network topology for subsequent status collection and collaborative analysis.

10. An integrated service system for intelligent security protection, used to implement the service method according to any one of claims 1 to 9, characterized in that: Includes data perception module, risk identification module, and collaborative control module; Data perception module: obtains information about all sensing devices and inspection devices in the protection area through the API interface of the protection area management platform, builds a multi-source sensing network for the sensing devices, and obtains park status data in real time; Risk identification module: After intelligently analyzing the park status data, it identifies the types of potential risk events, uses the spatiotemporal correlation algorithm to determine the risk propagation trend, uses the Transformer model to evaluate the risk level of each area in the park, and generates a risk level map; Collaborative control module: Path planning and scheduling optimization of inspection equipment are carried out according to the currently available inspection resources and risk level map, and the operation of all inspection equipment is collaboratively controlled based on the path planning and scheduling optimization results.

Images